1//! The `language` crate provides a large chunk of Zed's language-related
2//! features (the other big contributors being project and lsp crates that revolve around LSP features).
3//! Namely, this crate:
4//! - Provides [`Language`], [`Grammar`] and [`LanguageRegistry`] types that
5//! use Tree-sitter to provide syntax highlighting to the editor; note though that `language` doesn't perform the highlighting by itself. It only maps ranges in a buffer to colors. Treesitter is also used for buffer outlines (lists of symbols in a buffer)
6//! - Exposes [`LanguageConfig`] that describes how constructs (like brackets or line comments) should be handled by the editor for a source file of a particular language.
7//!
8//! Notably we do *not* assign a single language to a single file; in real world a single file can consist of multiple programming languages - HTML is a good example of that - and `language` crate tends to reflect that status quo in its API.
9mod buffer;
10mod diagnostic_set;
11mod highlight_map;
12mod language_registry;
13pub mod language_settings;
14mod outline;
15pub mod proto;
16mod syntax_map;
17mod task_context;
18
19#[cfg(test)]
20pub mod buffer_tests;
21pub mod markdown;
22
23use crate::language_settings::SoftWrap;
24use anyhow::{anyhow, Context, Result};
25use async_trait::async_trait;
26use collections::{HashMap, HashSet};
27use futures::Future;
28use gpui::{AppContext, AsyncAppContext, Model, SharedString, Task};
29pub use highlight_map::HighlightMap;
30use http_client::HttpClient;
31pub use language_registry::LanguageName;
32use lsp::{CodeActionKind, LanguageServerBinary};
33use parking_lot::Mutex;
34use regex::Regex;
35use schemars::{
36 gen::SchemaGenerator,
37 schema::{InstanceType, Schema, SchemaObject},
38 JsonSchema,
39};
40use serde::{de, Deserialize, Deserializer, Serialize, Serializer};
41use serde_json::Value;
42use settings::WorktreeId;
43use smol::future::FutureExt as _;
44use std::num::NonZeroU32;
45use std::{
46 any::Any,
47 ffi::OsStr,
48 fmt::Debug,
49 hash::Hash,
50 mem,
51 ops::{DerefMut, Range},
52 path::{Path, PathBuf},
53 pin::Pin,
54 str,
55 sync::{
56 atomic::{AtomicU64, AtomicUsize, Ordering::SeqCst},
57 Arc, LazyLock,
58 },
59};
60use syntax_map::{QueryCursorHandle, SyntaxSnapshot};
61use task::RunnableTag;
62pub use task_context::{ContextProvider, RunnableRange};
63use theme::SyntaxTheme;
64use tree_sitter::{self, wasmtime, Query, QueryCursor, WasmStore};
65use util::serde::default_true;
66
67pub use buffer::Operation;
68pub use buffer::*;
69pub use diagnostic_set::DiagnosticEntry;
70pub use language_registry::{
71 AvailableLanguage, LanguageNotFound, LanguageQueries, LanguageRegistry,
72 LanguageServerBinaryStatus, PendingLanguageServer, QUERY_FILENAME_PREFIXES,
73};
74pub use lsp::LanguageServerId;
75pub use outline::*;
76pub use syntax_map::{OwnedSyntaxLayer, SyntaxLayer};
77pub use text::{AnchorRangeExt, LineEnding};
78pub use tree_sitter::{Node, Parser, Tree, TreeCursor};
79
80/// Initializes the `language` crate.
81///
82/// This should be called before making use of items from the create.
83pub fn init(cx: &mut AppContext) {
84 language_settings::init(cx);
85}
86
87static QUERY_CURSORS: Mutex<Vec<QueryCursor>> = Mutex::new(vec![]);
88static PARSERS: Mutex<Vec<Parser>> = Mutex::new(vec![]);
89
90pub fn with_parser<F, R>(func: F) -> R
91where
92 F: FnOnce(&mut Parser) -> R,
93{
94 let mut parser = PARSERS.lock().pop().unwrap_or_else(|| {
95 let mut parser = Parser::new();
96 parser
97 .set_wasm_store(WasmStore::new(&WASM_ENGINE).unwrap())
98 .unwrap();
99 parser
100 });
101 parser.set_included_ranges(&[]).unwrap();
102 let result = func(&mut parser);
103 PARSERS.lock().push(parser);
104 result
105}
106
107pub fn with_query_cursor<F, R>(func: F) -> R
108where
109 F: FnOnce(&mut QueryCursor) -> R,
110{
111 let mut cursor = QueryCursorHandle::new();
112 func(cursor.deref_mut())
113}
114
115static NEXT_LANGUAGE_ID: LazyLock<AtomicUsize> = LazyLock::new(Default::default);
116static NEXT_GRAMMAR_ID: LazyLock<AtomicUsize> = LazyLock::new(Default::default);
117static WASM_ENGINE: LazyLock<wasmtime::Engine> = LazyLock::new(|| {
118 wasmtime::Engine::new(&wasmtime::Config::new()).expect("Failed to create Wasmtime engine")
119});
120
121/// A shared grammar for plain text, exposed for reuse by downstream crates.
122pub static PLAIN_TEXT: LazyLock<Arc<Language>> = LazyLock::new(|| {
123 Arc::new(Language::new(
124 LanguageConfig {
125 name: "Plain Text".into(),
126 soft_wrap: Some(SoftWrap::EditorWidth),
127 ..Default::default()
128 },
129 None,
130 ))
131});
132
133/// Types that represent a position in a buffer, and can be converted into
134/// an LSP position, to send to a language server.
135pub trait ToLspPosition {
136 /// Converts the value into an LSP position.
137 fn to_lsp_position(self) -> lsp::Position;
138}
139
140/// A name of a language server.
141#[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Deserialize, Serialize)]
142pub struct LanguageServerName(pub Arc<str>);
143
144impl LanguageServerName {
145 pub fn from_proto(s: String) -> Self {
146 Self(Arc::from(s))
147 }
148}
149
150#[derive(Debug, Clone, PartialEq, Eq, Hash)]
151pub struct Location {
152 pub buffer: Model<Buffer>,
153 pub range: Range<Anchor>,
154}
155
156/// Represents a Language Server, with certain cached sync properties.
157/// Uses [`LspAdapter`] under the hood, but calls all 'static' methods
158/// once at startup, and caches the results.
159pub struct CachedLspAdapter {
160 pub name: LanguageServerName,
161 pub disk_based_diagnostic_sources: Vec<String>,
162 pub disk_based_diagnostics_progress_token: Option<String>,
163 language_ids: HashMap<String, String>,
164 pub adapter: Arc<dyn LspAdapter>,
165 pub reinstall_attempt_count: AtomicU64,
166 cached_binary: futures::lock::Mutex<Option<LanguageServerBinary>>,
167}
168
169impl Debug for CachedLspAdapter {
170 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
171 f.debug_struct("CachedLspAdapter")
172 .field("name", &self.name)
173 .field(
174 "disk_based_diagnostic_sources",
175 &self.disk_based_diagnostic_sources,
176 )
177 .field(
178 "disk_based_diagnostics_progress_token",
179 &self.disk_based_diagnostics_progress_token,
180 )
181 .field("language_ids", &self.language_ids)
182 .field("reinstall_attempt_count", &self.reinstall_attempt_count)
183 .finish_non_exhaustive()
184 }
185}
186
187impl CachedLspAdapter {
188 pub fn new(adapter: Arc<dyn LspAdapter>) -> Arc<Self> {
189 let name = adapter.name();
190 let disk_based_diagnostic_sources = adapter.disk_based_diagnostic_sources();
191 let disk_based_diagnostics_progress_token = adapter.disk_based_diagnostics_progress_token();
192 let language_ids = adapter.language_ids();
193
194 Arc::new(CachedLspAdapter {
195 name,
196 disk_based_diagnostic_sources,
197 disk_based_diagnostics_progress_token,
198 language_ids,
199 adapter,
200 cached_binary: Default::default(),
201 reinstall_attempt_count: AtomicU64::new(0),
202 })
203 }
204
205 pub fn name(&self) -> Arc<str> {
206 self.adapter.name().0.clone()
207 }
208
209 pub async fn get_language_server_command(
210 self: Arc<Self>,
211 container_dir: Option<Arc<Path>>,
212 delegate: Arc<dyn LspAdapterDelegate>,
213 cx: &mut AsyncAppContext,
214 ) -> Result<LanguageServerBinary> {
215 let cached_binary = self.cached_binary.lock().await;
216 self.adapter
217 .clone()
218 .get_language_server_command(container_dir, delegate, cached_binary, cx)
219 .await
220 }
221
222 pub fn can_be_reinstalled(&self) -> bool {
223 self.adapter.can_be_reinstalled()
224 }
225
226 pub async fn installation_test_binary(
227 &self,
228 container_dir: PathBuf,
229 ) -> Option<LanguageServerBinary> {
230 self.adapter.installation_test_binary(container_dir).await
231 }
232
233 pub fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
234 self.adapter.code_action_kinds()
235 }
236
237 pub fn process_diagnostics(&self, params: &mut lsp::PublishDiagnosticsParams) {
238 self.adapter.process_diagnostics(params)
239 }
240
241 pub async fn process_completions(&self, completion_items: &mut [lsp::CompletionItem]) {
242 self.adapter.process_completions(completion_items).await
243 }
244
245 pub async fn labels_for_completions(
246 &self,
247 completion_items: &[lsp::CompletionItem],
248 language: &Arc<Language>,
249 ) -> Result<Vec<Option<CodeLabel>>> {
250 self.adapter
251 .clone()
252 .labels_for_completions(completion_items, language)
253 .await
254 }
255
256 pub async fn labels_for_symbols(
257 &self,
258 symbols: &[(String, lsp::SymbolKind)],
259 language: &Arc<Language>,
260 ) -> Result<Vec<Option<CodeLabel>>> {
261 self.adapter
262 .clone()
263 .labels_for_symbols(symbols, language)
264 .await
265 }
266
267 pub fn language_id(&self, language_name: &LanguageName) -> String {
268 self.language_ids
269 .get(language_name.0.as_ref())
270 .cloned()
271 .unwrap_or_else(|| language_name.lsp_id())
272 }
273}
274
275/// [`LspAdapterDelegate`] allows [`LspAdapter]` implementations to interface with the application
276// e.g. to display a notification or fetch data from the web.
277#[async_trait]
278pub trait LspAdapterDelegate: Send + Sync {
279 fn show_notification(&self, message: &str, cx: &mut AppContext);
280 fn http_client(&self) -> Arc<dyn HttpClient>;
281 fn worktree_id(&self) -> WorktreeId;
282 fn worktree_root_path(&self) -> &Path;
283 fn update_status(&self, language: LanguageServerName, status: LanguageServerBinaryStatus);
284
285 async fn which(&self, command: &OsStr) -> Option<PathBuf>;
286 async fn shell_env(&self) -> HashMap<String, String>;
287 async fn read_text_file(&self, path: PathBuf) -> Result<String>;
288 async fn try_exec(&self, binary: LanguageServerBinary) -> Result<()>;
289}
290
291#[async_trait(?Send)]
292pub trait LspAdapter: 'static + Send + Sync {
293 fn name(&self) -> LanguageServerName;
294
295 fn get_language_server_command<'a>(
296 self: Arc<Self>,
297 container_dir: Option<Arc<Path>>,
298 delegate: Arc<dyn LspAdapterDelegate>,
299 mut cached_binary: futures::lock::MutexGuard<'a, Option<LanguageServerBinary>>,
300 cx: &'a mut AsyncAppContext,
301 ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
302 async move {
303 // First we check whether the adapter can give us a user-installed binary.
304 // If so, we do *not* want to cache that, because each worktree might give us a different
305 // binary:
306 //
307 // worktree 1: user-installed at `.bin/gopls`
308 // worktree 2: user-installed at `~/bin/gopls`
309 // worktree 3: no gopls found in PATH -> fallback to Zed installation
310 //
311 // We only want to cache when we fall back to the global one,
312 // because we don't want to download and overwrite our global one
313 // for each worktree we might have open.
314 if let Some(binary) = self.check_if_user_installed(delegate.as_ref(), cx).await {
315 log::info!(
316 "found user-installed language server for {}. path: {:?}, arguments: {:?}",
317 self.name().0,
318 binary.path,
319 binary.arguments
320 );
321 return Ok(binary);
322 }
323
324 if let Some(cached_binary) = cached_binary.as_ref() {
325 return Ok(cached_binary.clone());
326 }
327
328 let Some(container_dir) = container_dir else {
329 anyhow::bail!("cannot download language servers for remotes (yet)")
330 };
331
332 if !container_dir.exists() {
333 smol::fs::create_dir_all(&container_dir)
334 .await
335 .context("failed to create container directory")?;
336 }
337
338 let mut binary = try_fetch_server_binary(self.as_ref(), &delegate, container_dir.to_path_buf(), cx).await;
339
340 if let Err(error) = binary.as_ref() {
341 if let Some(prev_downloaded_binary) = self
342 .cached_server_binary(container_dir.to_path_buf(), delegate.as_ref())
343 .await
344 {
345 log::info!(
346 "failed to fetch newest version of language server {:?}. falling back to using {:?}",
347 self.name(),
348 prev_downloaded_binary.path
349 );
350 binary = Ok(prev_downloaded_binary);
351 } else {
352 delegate.update_status(
353 self.name(),
354 LanguageServerBinaryStatus::Failed {
355 error: format!("{error:?}"),
356 },
357 );
358 }
359 }
360
361 if let Ok(binary) = &binary {
362 *cached_binary = Some(binary.clone());
363 }
364
365 binary
366 }
367 .boxed_local()
368 }
369
370 async fn check_if_user_installed(
371 &self,
372 _: &dyn LspAdapterDelegate,
373 _: &AsyncAppContext,
374 ) -> Option<LanguageServerBinary> {
375 None
376 }
377
378 async fn fetch_latest_server_version(
379 &self,
380 delegate: &dyn LspAdapterDelegate,
381 ) -> Result<Box<dyn 'static + Send + Any>>;
382
383 fn will_fetch_server(
384 &self,
385 _: &Arc<dyn LspAdapterDelegate>,
386 _: &mut AsyncAppContext,
387 ) -> Option<Task<Result<()>>> {
388 None
389 }
390
391 async fn fetch_server_binary(
392 &self,
393 latest_version: Box<dyn 'static + Send + Any>,
394 container_dir: PathBuf,
395 delegate: &dyn LspAdapterDelegate,
396 ) -> Result<LanguageServerBinary>;
397
398 async fn cached_server_binary(
399 &self,
400 container_dir: PathBuf,
401 delegate: &dyn LspAdapterDelegate,
402 ) -> Option<LanguageServerBinary>;
403
404 /// Returns `true` if a language server can be reinstalled.
405 ///
406 /// If language server initialization fails, a reinstallation will be attempted unless the value returned from this method is `false`.
407 ///
408 /// Implementations that rely on software already installed on user's system
409 /// should have [`can_be_reinstalled`](Self::can_be_reinstalled) return `false`.
410 fn can_be_reinstalled(&self) -> bool {
411 true
412 }
413
414 async fn installation_test_binary(
415 &self,
416 container_dir: PathBuf,
417 ) -> Option<LanguageServerBinary>;
418
419 fn process_diagnostics(&self, _: &mut lsp::PublishDiagnosticsParams) {}
420
421 /// Post-processes completions provided by the language server.
422 async fn process_completions(&self, _: &mut [lsp::CompletionItem]) {}
423
424 async fn labels_for_completions(
425 self: Arc<Self>,
426 completions: &[lsp::CompletionItem],
427 language: &Arc<Language>,
428 ) -> Result<Vec<Option<CodeLabel>>> {
429 let mut labels = Vec::new();
430 for (ix, completion) in completions.iter().enumerate() {
431 let label = self.label_for_completion(completion, language).await;
432 if let Some(label) = label {
433 labels.resize(ix + 1, None);
434 *labels.last_mut().unwrap() = Some(label);
435 }
436 }
437 Ok(labels)
438 }
439
440 async fn label_for_completion(
441 &self,
442 _: &lsp::CompletionItem,
443 _: &Arc<Language>,
444 ) -> Option<CodeLabel> {
445 None
446 }
447
448 async fn labels_for_symbols(
449 self: Arc<Self>,
450 symbols: &[(String, lsp::SymbolKind)],
451 language: &Arc<Language>,
452 ) -> Result<Vec<Option<CodeLabel>>> {
453 let mut labels = Vec::new();
454 for (ix, (name, kind)) in symbols.iter().enumerate() {
455 let label = self.label_for_symbol(name, *kind, language).await;
456 if let Some(label) = label {
457 labels.resize(ix + 1, None);
458 *labels.last_mut().unwrap() = Some(label);
459 }
460 }
461 Ok(labels)
462 }
463
464 async fn label_for_symbol(
465 &self,
466 _: &str,
467 _: lsp::SymbolKind,
468 _: &Arc<Language>,
469 ) -> Option<CodeLabel> {
470 None
471 }
472
473 /// Returns initialization options that are going to be sent to a LSP server as a part of [`lsp::InitializeParams`]
474 async fn initialization_options(
475 self: Arc<Self>,
476 _: &Arc<dyn LspAdapterDelegate>,
477 ) -> Result<Option<Value>> {
478 Ok(None)
479 }
480
481 async fn workspace_configuration(
482 self: Arc<Self>,
483 _: &Arc<dyn LspAdapterDelegate>,
484 _cx: &mut AsyncAppContext,
485 ) -> Result<Value> {
486 Ok(serde_json::json!({}))
487 }
488
489 /// Returns a list of code actions supported by a given LspAdapter
490 fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
491 Some(vec![
492 CodeActionKind::EMPTY,
493 CodeActionKind::QUICKFIX,
494 CodeActionKind::REFACTOR,
495 CodeActionKind::REFACTOR_EXTRACT,
496 CodeActionKind::SOURCE,
497 ])
498 }
499
500 fn disk_based_diagnostic_sources(&self) -> Vec<String> {
501 Default::default()
502 }
503
504 fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
505 None
506 }
507
508 fn language_ids(&self) -> HashMap<String, String> {
509 Default::default()
510 }
511}
512
513async fn try_fetch_server_binary<L: LspAdapter + 'static + Send + Sync + ?Sized>(
514 adapter: &L,
515 delegate: &Arc<dyn LspAdapterDelegate>,
516 container_dir: PathBuf,
517 cx: &mut AsyncAppContext,
518) -> Result<LanguageServerBinary> {
519 if let Some(task) = adapter.will_fetch_server(delegate, cx) {
520 task.await?;
521 }
522
523 let name = adapter.name();
524 log::info!("fetching latest version of language server {:?}", name.0);
525 delegate.update_status(name.clone(), LanguageServerBinaryStatus::CheckingForUpdate);
526 let latest_version = adapter
527 .fetch_latest_server_version(delegate.as_ref())
528 .await?;
529
530 log::info!("downloading language server {:?}", name.0);
531 delegate.update_status(adapter.name(), LanguageServerBinaryStatus::Downloading);
532 let binary = adapter
533 .fetch_server_binary(latest_version, container_dir, delegate.as_ref())
534 .await;
535
536 delegate.update_status(name.clone(), LanguageServerBinaryStatus::None);
537 binary
538}
539
540#[derive(Clone, Debug, Default, PartialEq, Eq)]
541pub struct CodeLabel {
542 /// The text to display.
543 pub text: String,
544 /// Syntax highlighting runs.
545 pub runs: Vec<(Range<usize>, HighlightId)>,
546 /// The portion of the text that should be used in fuzzy filtering.
547 pub filter_range: Range<usize>,
548}
549
550#[derive(Clone, Deserialize, JsonSchema)]
551pub struct LanguageConfig {
552 /// Human-readable name of the language.
553 pub name: LanguageName,
554 /// The name of this language for a Markdown code fence block
555 pub code_fence_block_name: Option<Arc<str>>,
556 // The name of the grammar in a WASM bundle (experimental).
557 pub grammar: Option<Arc<str>>,
558 /// The criteria for matching this language to a given file.
559 #[serde(flatten)]
560 pub matcher: LanguageMatcher,
561 /// List of bracket types in a language.
562 #[serde(default)]
563 #[schemars(schema_with = "bracket_pair_config_json_schema")]
564 pub brackets: BracketPairConfig,
565 /// If set to true, auto indentation uses last non empty line to determine
566 /// the indentation level for a new line.
567 #[serde(default = "auto_indent_using_last_non_empty_line_default")]
568 pub auto_indent_using_last_non_empty_line: bool,
569 /// A regex that is used to determine whether the indentation level should be
570 /// increased in the following line.
571 #[serde(default, deserialize_with = "deserialize_regex")]
572 #[schemars(schema_with = "regex_json_schema")]
573 pub increase_indent_pattern: Option<Regex>,
574 /// A regex that is used to determine whether the indentation level should be
575 /// decreased in the following line.
576 #[serde(default, deserialize_with = "deserialize_regex")]
577 #[schemars(schema_with = "regex_json_schema")]
578 pub decrease_indent_pattern: Option<Regex>,
579 /// A list of characters that trigger the automatic insertion of a closing
580 /// bracket when they immediately precede the point where an opening
581 /// bracket is inserted.
582 #[serde(default)]
583 pub autoclose_before: String,
584 /// A placeholder used internally by Semantic Index.
585 #[serde(default)]
586 pub collapsed_placeholder: String,
587 /// A line comment string that is inserted in e.g. `toggle comments` action.
588 /// A language can have multiple flavours of line comments. All of the provided line comments are
589 /// used for comment continuations on the next line, but only the first one is used for Editor::ToggleComments.
590 #[serde(default)]
591 pub line_comments: Vec<Arc<str>>,
592 /// Starting and closing characters of a block comment.
593 #[serde(default)]
594 pub block_comment: Option<(Arc<str>, Arc<str>)>,
595 /// A list of language servers that are allowed to run on subranges of a given language.
596 #[serde(default)]
597 pub scope_opt_in_language_servers: Vec<String>,
598 #[serde(default)]
599 pub overrides: HashMap<String, LanguageConfigOverride>,
600 /// A list of characters that Zed should treat as word characters for the
601 /// purpose of features that operate on word boundaries, like 'move to next word end'
602 /// or a whole-word search in buffer search.
603 #[serde(default)]
604 pub word_characters: HashSet<char>,
605 /// Whether to indent lines using tab characters, as opposed to multiple
606 /// spaces.
607 #[serde(default)]
608 pub hard_tabs: Option<bool>,
609 /// How many columns a tab should occupy.
610 #[serde(default)]
611 pub tab_size: Option<NonZeroU32>,
612 /// How to soft-wrap long lines of text.
613 #[serde(default)]
614 pub soft_wrap: Option<SoftWrap>,
615 /// The name of a Prettier parser that will be used for this language when no file path is available.
616 /// If there's a parser name in the language settings, that will be used instead.
617 #[serde(default)]
618 pub prettier_parser_name: Option<String>,
619 /// If true, this language is only for syntax highlighting via an injection into other
620 /// languages, but should not appear to the user as a distinct language.
621 #[serde(default)]
622 pub hidden: bool,
623}
624
625#[derive(Clone, Debug, Serialize, Deserialize, Default, JsonSchema)]
626pub struct LanguageMatcher {
627 /// Given a list of `LanguageConfig`'s, the language of a file can be determined based on the path extension matching any of the `path_suffixes`.
628 #[serde(default)]
629 pub path_suffixes: Vec<String>,
630 /// A regex pattern that determines whether the language should be assigned to a file or not.
631 #[serde(
632 default,
633 serialize_with = "serialize_regex",
634 deserialize_with = "deserialize_regex"
635 )]
636 #[schemars(schema_with = "regex_json_schema")]
637 pub first_line_pattern: Option<Regex>,
638}
639
640/// Represents a language for the given range. Some languages (e.g. HTML)
641/// interleave several languages together, thus a single buffer might actually contain
642/// several nested scopes.
643#[derive(Clone, Debug)]
644pub struct LanguageScope {
645 language: Arc<Language>,
646 override_id: Option<u32>,
647}
648
649#[derive(Clone, Deserialize, Default, Debug, JsonSchema)]
650pub struct LanguageConfigOverride {
651 #[serde(default)]
652 pub line_comments: Override<Vec<Arc<str>>>,
653 #[serde(default)]
654 pub block_comment: Override<(Arc<str>, Arc<str>)>,
655 #[serde(skip_deserializing)]
656 #[schemars(skip)]
657 pub disabled_bracket_ixs: Vec<u16>,
658 #[serde(default)]
659 pub word_characters: Override<HashSet<char>>,
660 #[serde(default)]
661 pub opt_into_language_servers: Vec<String>,
662}
663
664#[derive(Clone, Deserialize, Debug, Serialize, JsonSchema)]
665#[serde(untagged)]
666pub enum Override<T> {
667 Remove { remove: bool },
668 Set(T),
669}
670
671impl<T> Default for Override<T> {
672 fn default() -> Self {
673 Override::Remove { remove: false }
674 }
675}
676
677impl<T> Override<T> {
678 fn as_option<'a>(this: Option<&'a Self>, original: Option<&'a T>) -> Option<&'a T> {
679 match this {
680 Some(Self::Set(value)) => Some(value),
681 Some(Self::Remove { remove: true }) => None,
682 Some(Self::Remove { remove: false }) | None => original,
683 }
684 }
685}
686
687impl Default for LanguageConfig {
688 fn default() -> Self {
689 Self {
690 name: LanguageName::new(""),
691 code_fence_block_name: None,
692 grammar: None,
693 matcher: LanguageMatcher::default(),
694 brackets: Default::default(),
695 auto_indent_using_last_non_empty_line: auto_indent_using_last_non_empty_line_default(),
696 increase_indent_pattern: Default::default(),
697 decrease_indent_pattern: Default::default(),
698 autoclose_before: Default::default(),
699 line_comments: Default::default(),
700 block_comment: Default::default(),
701 scope_opt_in_language_servers: Default::default(),
702 overrides: Default::default(),
703 word_characters: Default::default(),
704 collapsed_placeholder: Default::default(),
705 hard_tabs: None,
706 tab_size: None,
707 soft_wrap: None,
708 prettier_parser_name: None,
709 hidden: false,
710 }
711 }
712}
713
714fn auto_indent_using_last_non_empty_line_default() -> bool {
715 true
716}
717
718fn deserialize_regex<'de, D: Deserializer<'de>>(d: D) -> Result<Option<Regex>, D::Error> {
719 let source = Option::<String>::deserialize(d)?;
720 if let Some(source) = source {
721 Ok(Some(regex::Regex::new(&source).map_err(de::Error::custom)?))
722 } else {
723 Ok(None)
724 }
725}
726
727fn regex_json_schema(_: &mut SchemaGenerator) -> Schema {
728 Schema::Object(SchemaObject {
729 instance_type: Some(InstanceType::String.into()),
730 ..Default::default()
731 })
732}
733
734fn serialize_regex<S>(regex: &Option<Regex>, serializer: S) -> Result<S::Ok, S::Error>
735where
736 S: Serializer,
737{
738 match regex {
739 Some(regex) => serializer.serialize_str(regex.as_str()),
740 None => serializer.serialize_none(),
741 }
742}
743
744#[doc(hidden)]
745#[cfg(any(test, feature = "test-support"))]
746pub struct FakeLspAdapter {
747 pub name: &'static str,
748 pub initialization_options: Option<Value>,
749 pub prettier_plugins: Vec<&'static str>,
750 pub disk_based_diagnostics_progress_token: Option<String>,
751 pub disk_based_diagnostics_sources: Vec<String>,
752 pub language_server_binary: LanguageServerBinary,
753
754 pub capabilities: lsp::ServerCapabilities,
755 pub initializer: Option<Box<dyn 'static + Send + Sync + Fn(&mut lsp::FakeLanguageServer)>>,
756}
757
758/// Configuration of handling bracket pairs for a given language.
759///
760/// This struct includes settings for defining which pairs of characters are considered brackets and
761/// also specifies any language-specific scopes where these pairs should be ignored for bracket matching purposes.
762#[derive(Clone, Debug, Default, JsonSchema)]
763pub struct BracketPairConfig {
764 /// A list of character pairs that should be treated as brackets in the context of a given language.
765 pub pairs: Vec<BracketPair>,
766 /// A list of tree-sitter scopes for which a given bracket should not be active.
767 /// N-th entry in `[Self::disabled_scopes_by_bracket_ix]` contains a list of disabled scopes for an n-th entry in `[Self::pairs]`
768 #[schemars(skip)]
769 pub disabled_scopes_by_bracket_ix: Vec<Vec<String>>,
770}
771
772fn bracket_pair_config_json_schema(gen: &mut SchemaGenerator) -> Schema {
773 Option::<Vec<BracketPairContent>>::json_schema(gen)
774}
775
776#[derive(Deserialize, JsonSchema)]
777pub struct BracketPairContent {
778 #[serde(flatten)]
779 pub bracket_pair: BracketPair,
780 #[serde(default)]
781 pub not_in: Vec<String>,
782}
783
784impl<'de> Deserialize<'de> for BracketPairConfig {
785 fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
786 where
787 D: Deserializer<'de>,
788 {
789 let result = Vec::<BracketPairContent>::deserialize(deserializer)?;
790 let mut brackets = Vec::with_capacity(result.len());
791 let mut disabled_scopes_by_bracket_ix = Vec::with_capacity(result.len());
792 for entry in result {
793 brackets.push(entry.bracket_pair);
794 disabled_scopes_by_bracket_ix.push(entry.not_in);
795 }
796
797 Ok(BracketPairConfig {
798 pairs: brackets,
799 disabled_scopes_by_bracket_ix,
800 })
801 }
802}
803
804/// Describes a single bracket pair and how an editor should react to e.g. inserting
805/// an opening bracket or to a newline character insertion in between `start` and `end` characters.
806#[derive(Clone, Debug, Default, Deserialize, PartialEq, JsonSchema)]
807pub struct BracketPair {
808 /// Starting substring for a bracket.
809 pub start: String,
810 /// Ending substring for a bracket.
811 pub end: String,
812 /// True if `end` should be automatically inserted right after `start` characters.
813 pub close: bool,
814 /// True if selected text should be surrounded by `start` and `end` characters.
815 #[serde(default = "default_true")]
816 pub surround: bool,
817 /// True if an extra newline should be inserted while the cursor is in the middle
818 /// of that bracket pair.
819 pub newline: bool,
820}
821
822#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
823pub(crate) struct LanguageId(usize);
824
825impl LanguageId {
826 pub(crate) fn new() -> Self {
827 Self(NEXT_LANGUAGE_ID.fetch_add(1, SeqCst))
828 }
829}
830
831pub struct Language {
832 pub(crate) id: LanguageId,
833 pub(crate) config: LanguageConfig,
834 pub(crate) grammar: Option<Arc<Grammar>>,
835 pub(crate) context_provider: Option<Arc<dyn ContextProvider>>,
836}
837
838#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
839pub struct GrammarId(pub usize);
840
841impl GrammarId {
842 pub(crate) fn new() -> Self {
843 Self(NEXT_GRAMMAR_ID.fetch_add(1, SeqCst))
844 }
845}
846
847pub struct Grammar {
848 id: GrammarId,
849 pub ts_language: tree_sitter::Language,
850 pub(crate) error_query: Query,
851 pub(crate) highlights_query: Option<Query>,
852 pub(crate) brackets_config: Option<BracketConfig>,
853 pub(crate) redactions_config: Option<RedactionConfig>,
854 pub(crate) runnable_config: Option<RunnableConfig>,
855 pub(crate) indents_config: Option<IndentConfig>,
856 pub outline_config: Option<OutlineConfig>,
857 pub embedding_config: Option<EmbeddingConfig>,
858 pub(crate) injection_config: Option<InjectionConfig>,
859 pub(crate) override_config: Option<OverrideConfig>,
860 pub(crate) highlight_map: Mutex<HighlightMap>,
861}
862
863struct IndentConfig {
864 query: Query,
865 indent_capture_ix: u32,
866 start_capture_ix: Option<u32>,
867 end_capture_ix: Option<u32>,
868 outdent_capture_ix: Option<u32>,
869}
870
871pub struct OutlineConfig {
872 pub query: Query,
873 pub item_capture_ix: u32,
874 pub name_capture_ix: u32,
875 pub context_capture_ix: Option<u32>,
876 pub extra_context_capture_ix: Option<u32>,
877 pub open_capture_ix: Option<u32>,
878 pub close_capture_ix: Option<u32>,
879 pub annotation_capture_ix: Option<u32>,
880}
881
882#[derive(Debug)]
883pub struct EmbeddingConfig {
884 pub query: Query,
885 pub item_capture_ix: u32,
886 pub name_capture_ix: Option<u32>,
887 pub context_capture_ix: Option<u32>,
888 pub collapse_capture_ix: Option<u32>,
889 pub keep_capture_ix: Option<u32>,
890}
891
892struct InjectionConfig {
893 query: Query,
894 content_capture_ix: u32,
895 language_capture_ix: Option<u32>,
896 patterns: Vec<InjectionPatternConfig>,
897}
898
899struct RedactionConfig {
900 pub query: Query,
901 pub redaction_capture_ix: u32,
902}
903
904#[derive(Clone, Debug, PartialEq)]
905enum RunnableCapture {
906 Named(SharedString),
907 Run,
908}
909
910struct RunnableConfig {
911 pub query: Query,
912 /// A mapping from capture indice to capture kind
913 pub extra_captures: Vec<RunnableCapture>,
914}
915
916struct OverrideConfig {
917 query: Query,
918 values: HashMap<u32, (String, LanguageConfigOverride)>,
919}
920
921#[derive(Default, Clone)]
922struct InjectionPatternConfig {
923 language: Option<Box<str>>,
924 combined: bool,
925}
926
927struct BracketConfig {
928 query: Query,
929 open_capture_ix: u32,
930 close_capture_ix: u32,
931}
932
933impl Language {
934 pub fn new(config: LanguageConfig, ts_language: Option<tree_sitter::Language>) -> Self {
935 Self::new_with_id(LanguageId::new(), config, ts_language)
936 }
937
938 fn new_with_id(
939 id: LanguageId,
940 config: LanguageConfig,
941 ts_language: Option<tree_sitter::Language>,
942 ) -> Self {
943 Self {
944 id,
945 config,
946 grammar: ts_language.map(|ts_language| {
947 Arc::new(Grammar {
948 id: GrammarId::new(),
949 highlights_query: None,
950 brackets_config: None,
951 outline_config: None,
952 embedding_config: None,
953 indents_config: None,
954 injection_config: None,
955 override_config: None,
956 redactions_config: None,
957 runnable_config: None,
958 error_query: Query::new(&ts_language, "(ERROR) @error").unwrap(),
959 ts_language,
960 highlight_map: Default::default(),
961 })
962 }),
963 context_provider: None,
964 }
965 }
966
967 pub fn with_context_provider(mut self, provider: Option<Arc<dyn ContextProvider>>) -> Self {
968 self.context_provider = provider;
969 self
970 }
971
972 pub fn with_queries(mut self, queries: LanguageQueries) -> Result<Self> {
973 if let Some(query) = queries.highlights {
974 self = self
975 .with_highlights_query(query.as_ref())
976 .context("Error loading highlights query")?;
977 }
978 if let Some(query) = queries.brackets {
979 self = self
980 .with_brackets_query(query.as_ref())
981 .context("Error loading brackets query")?;
982 }
983 if let Some(query) = queries.indents {
984 self = self
985 .with_indents_query(query.as_ref())
986 .context("Error loading indents query")?;
987 }
988 if let Some(query) = queries.outline {
989 self = self
990 .with_outline_query(query.as_ref())
991 .context("Error loading outline query")?;
992 }
993 if let Some(query) = queries.embedding {
994 self = self
995 .with_embedding_query(query.as_ref())
996 .context("Error loading embedding query")?;
997 }
998 if let Some(query) = queries.injections {
999 self = self
1000 .with_injection_query(query.as_ref())
1001 .context("Error loading injection query")?;
1002 }
1003 if let Some(query) = queries.overrides {
1004 self = self
1005 .with_override_query(query.as_ref())
1006 .context("Error loading override query")?;
1007 }
1008 if let Some(query) = queries.redactions {
1009 self = self
1010 .with_redaction_query(query.as_ref())
1011 .context("Error loading redaction query")?;
1012 }
1013 if let Some(query) = queries.runnables {
1014 self = self
1015 .with_runnable_query(query.as_ref())
1016 .context("Error loading tests query")?;
1017 }
1018 Ok(self)
1019 }
1020
1021 pub fn with_highlights_query(mut self, source: &str) -> Result<Self> {
1022 let grammar = self
1023 .grammar_mut()
1024 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1025 grammar.highlights_query = Some(Query::new(&grammar.ts_language, source)?);
1026 Ok(self)
1027 }
1028
1029 pub fn with_runnable_query(mut self, source: &str) -> Result<Self> {
1030 let grammar = self
1031 .grammar_mut()
1032 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1033
1034 let query = Query::new(&grammar.ts_language, source)?;
1035 let mut extra_captures = Vec::with_capacity(query.capture_names().len());
1036
1037 for name in query.capture_names().iter() {
1038 let kind = if *name == "run" {
1039 RunnableCapture::Run
1040 } else {
1041 RunnableCapture::Named(name.to_string().into())
1042 };
1043 extra_captures.push(kind);
1044 }
1045
1046 grammar.runnable_config = Some(RunnableConfig {
1047 extra_captures,
1048 query,
1049 });
1050
1051 Ok(self)
1052 }
1053
1054 pub fn with_outline_query(mut self, source: &str) -> Result<Self> {
1055 let grammar = self
1056 .grammar_mut()
1057 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1058 let query = Query::new(&grammar.ts_language, source)?;
1059 let mut item_capture_ix = None;
1060 let mut name_capture_ix = None;
1061 let mut context_capture_ix = None;
1062 let mut extra_context_capture_ix = None;
1063 let mut open_capture_ix = None;
1064 let mut close_capture_ix = None;
1065 let mut annotation_capture_ix = None;
1066 get_capture_indices(
1067 &query,
1068 &mut [
1069 ("item", &mut item_capture_ix),
1070 ("name", &mut name_capture_ix),
1071 ("context", &mut context_capture_ix),
1072 ("context.extra", &mut extra_context_capture_ix),
1073 ("open", &mut open_capture_ix),
1074 ("close", &mut close_capture_ix),
1075 ("annotation", &mut annotation_capture_ix),
1076 ],
1077 );
1078 if let Some((item_capture_ix, name_capture_ix)) = item_capture_ix.zip(name_capture_ix) {
1079 grammar.outline_config = Some(OutlineConfig {
1080 query,
1081 item_capture_ix,
1082 name_capture_ix,
1083 context_capture_ix,
1084 extra_context_capture_ix,
1085 open_capture_ix,
1086 close_capture_ix,
1087 annotation_capture_ix,
1088 });
1089 }
1090 Ok(self)
1091 }
1092
1093 pub fn with_embedding_query(mut self, source: &str) -> Result<Self> {
1094 let grammar = self
1095 .grammar_mut()
1096 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1097 let query = Query::new(&grammar.ts_language, source)?;
1098 let mut item_capture_ix = None;
1099 let mut name_capture_ix = None;
1100 let mut context_capture_ix = None;
1101 let mut collapse_capture_ix = None;
1102 let mut keep_capture_ix = None;
1103 get_capture_indices(
1104 &query,
1105 &mut [
1106 ("item", &mut item_capture_ix),
1107 ("name", &mut name_capture_ix),
1108 ("context", &mut context_capture_ix),
1109 ("keep", &mut keep_capture_ix),
1110 ("collapse", &mut collapse_capture_ix),
1111 ],
1112 );
1113 if let Some(item_capture_ix) = item_capture_ix {
1114 grammar.embedding_config = Some(EmbeddingConfig {
1115 query,
1116 item_capture_ix,
1117 name_capture_ix,
1118 context_capture_ix,
1119 collapse_capture_ix,
1120 keep_capture_ix,
1121 });
1122 }
1123 Ok(self)
1124 }
1125
1126 pub fn with_brackets_query(mut self, source: &str) -> Result<Self> {
1127 let grammar = self
1128 .grammar_mut()
1129 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1130 let query = Query::new(&grammar.ts_language, source)?;
1131 let mut open_capture_ix = None;
1132 let mut close_capture_ix = None;
1133 get_capture_indices(
1134 &query,
1135 &mut [
1136 ("open", &mut open_capture_ix),
1137 ("close", &mut close_capture_ix),
1138 ],
1139 );
1140 if let Some((open_capture_ix, close_capture_ix)) = open_capture_ix.zip(close_capture_ix) {
1141 grammar.brackets_config = Some(BracketConfig {
1142 query,
1143 open_capture_ix,
1144 close_capture_ix,
1145 });
1146 }
1147 Ok(self)
1148 }
1149
1150 pub fn with_indents_query(mut self, source: &str) -> Result<Self> {
1151 let grammar = self
1152 .grammar_mut()
1153 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1154 let query = Query::new(&grammar.ts_language, source)?;
1155 let mut indent_capture_ix = None;
1156 let mut start_capture_ix = None;
1157 let mut end_capture_ix = None;
1158 let mut outdent_capture_ix = None;
1159 get_capture_indices(
1160 &query,
1161 &mut [
1162 ("indent", &mut indent_capture_ix),
1163 ("start", &mut start_capture_ix),
1164 ("end", &mut end_capture_ix),
1165 ("outdent", &mut outdent_capture_ix),
1166 ],
1167 );
1168 if let Some(indent_capture_ix) = indent_capture_ix {
1169 grammar.indents_config = Some(IndentConfig {
1170 query,
1171 indent_capture_ix,
1172 start_capture_ix,
1173 end_capture_ix,
1174 outdent_capture_ix,
1175 });
1176 }
1177 Ok(self)
1178 }
1179
1180 pub fn with_injection_query(mut self, source: &str) -> Result<Self> {
1181 let grammar = self
1182 .grammar_mut()
1183 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1184 let query = Query::new(&grammar.ts_language, source)?;
1185 let mut language_capture_ix = None;
1186 let mut content_capture_ix = None;
1187 get_capture_indices(
1188 &query,
1189 &mut [
1190 ("language", &mut language_capture_ix),
1191 ("content", &mut content_capture_ix),
1192 ],
1193 );
1194 let patterns = (0..query.pattern_count())
1195 .map(|ix| {
1196 let mut config = InjectionPatternConfig::default();
1197 for setting in query.property_settings(ix) {
1198 match setting.key.as_ref() {
1199 "language" => {
1200 config.language.clone_from(&setting.value);
1201 }
1202 "combined" => {
1203 config.combined = true;
1204 }
1205 _ => {}
1206 }
1207 }
1208 config
1209 })
1210 .collect();
1211 if let Some(content_capture_ix) = content_capture_ix {
1212 grammar.injection_config = Some(InjectionConfig {
1213 query,
1214 language_capture_ix,
1215 content_capture_ix,
1216 patterns,
1217 });
1218 }
1219 Ok(self)
1220 }
1221
1222 pub fn with_override_query(mut self, source: &str) -> anyhow::Result<Self> {
1223 let query = {
1224 let grammar = self
1225 .grammar
1226 .as_ref()
1227 .ok_or_else(|| anyhow!("no grammar for language"))?;
1228 Query::new(&grammar.ts_language, source)?
1229 };
1230
1231 let mut override_configs_by_id = HashMap::default();
1232 for (ix, name) in query.capture_names().iter().enumerate() {
1233 if !name.starts_with('_') {
1234 let value = self.config.overrides.remove(*name).unwrap_or_default();
1235 for server_name in &value.opt_into_language_servers {
1236 if !self
1237 .config
1238 .scope_opt_in_language_servers
1239 .contains(server_name)
1240 {
1241 util::debug_panic!("Server {server_name:?} has been opted-in by scope {name:?} but has not been marked as an opt-in server");
1242 }
1243 }
1244
1245 override_configs_by_id.insert(ix as u32, (name.to_string(), value));
1246 }
1247 }
1248
1249 if !self.config.overrides.is_empty() {
1250 let keys = self.config.overrides.keys().collect::<Vec<_>>();
1251 Err(anyhow!(
1252 "language {:?} has overrides in config not in query: {keys:?}",
1253 self.config.name
1254 ))?;
1255 }
1256
1257 for disabled_scope_name in self
1258 .config
1259 .brackets
1260 .disabled_scopes_by_bracket_ix
1261 .iter()
1262 .flatten()
1263 {
1264 if !override_configs_by_id
1265 .values()
1266 .any(|(scope_name, _)| scope_name == disabled_scope_name)
1267 {
1268 Err(anyhow!(
1269 "language {:?} has overrides in config not in query: {disabled_scope_name:?}",
1270 self.config.name
1271 ))?;
1272 }
1273 }
1274
1275 for (name, override_config) in override_configs_by_id.values_mut() {
1276 override_config.disabled_bracket_ixs = self
1277 .config
1278 .brackets
1279 .disabled_scopes_by_bracket_ix
1280 .iter()
1281 .enumerate()
1282 .filter_map(|(ix, disabled_scope_names)| {
1283 if disabled_scope_names.contains(name) {
1284 Some(ix as u16)
1285 } else {
1286 None
1287 }
1288 })
1289 .collect();
1290 }
1291
1292 self.config.brackets.disabled_scopes_by_bracket_ix.clear();
1293
1294 let grammar = self
1295 .grammar_mut()
1296 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1297 grammar.override_config = Some(OverrideConfig {
1298 query,
1299 values: override_configs_by_id,
1300 });
1301 Ok(self)
1302 }
1303
1304 pub fn with_redaction_query(mut self, source: &str) -> anyhow::Result<Self> {
1305 let grammar = self
1306 .grammar_mut()
1307 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1308
1309 let query = Query::new(&grammar.ts_language, source)?;
1310 let mut redaction_capture_ix = None;
1311 get_capture_indices(&query, &mut [("redact", &mut redaction_capture_ix)]);
1312
1313 if let Some(redaction_capture_ix) = redaction_capture_ix {
1314 grammar.redactions_config = Some(RedactionConfig {
1315 query,
1316 redaction_capture_ix,
1317 });
1318 }
1319
1320 Ok(self)
1321 }
1322
1323 fn grammar_mut(&mut self) -> Option<&mut Grammar> {
1324 Arc::get_mut(self.grammar.as_mut()?)
1325 }
1326
1327 pub fn name(&self) -> LanguageName {
1328 self.config.name.clone()
1329 }
1330
1331 pub fn code_fence_block_name(&self) -> Arc<str> {
1332 self.config
1333 .code_fence_block_name
1334 .clone()
1335 .unwrap_or_else(|| self.config.name.0.to_lowercase().into())
1336 }
1337
1338 pub fn context_provider(&self) -> Option<Arc<dyn ContextProvider>> {
1339 self.context_provider.clone()
1340 }
1341
1342 pub fn highlight_text<'a>(
1343 self: &'a Arc<Self>,
1344 text: &'a Rope,
1345 range: Range<usize>,
1346 ) -> Vec<(Range<usize>, HighlightId)> {
1347 let mut result = Vec::new();
1348 if let Some(grammar) = &self.grammar {
1349 let tree = grammar.parse_text(text, None);
1350 let captures =
1351 SyntaxSnapshot::single_tree_captures(range.clone(), text, &tree, self, |grammar| {
1352 grammar.highlights_query.as_ref()
1353 });
1354 let highlight_maps = vec![grammar.highlight_map()];
1355 let mut offset = 0;
1356 for chunk in
1357 BufferChunks::new(text, range, Some((captures, highlight_maps)), false, None)
1358 {
1359 let end_offset = offset + chunk.text.len();
1360 if let Some(highlight_id) = chunk.syntax_highlight_id {
1361 if !highlight_id.is_default() {
1362 result.push((offset..end_offset, highlight_id));
1363 }
1364 }
1365 offset = end_offset;
1366 }
1367 }
1368 result
1369 }
1370
1371 pub fn path_suffixes(&self) -> &[String] {
1372 &self.config.matcher.path_suffixes
1373 }
1374
1375 pub fn should_autoclose_before(&self, c: char) -> bool {
1376 c.is_whitespace() || self.config.autoclose_before.contains(c)
1377 }
1378
1379 pub fn set_theme(&self, theme: &SyntaxTheme) {
1380 if let Some(grammar) = self.grammar.as_ref() {
1381 if let Some(highlights_query) = &grammar.highlights_query {
1382 *grammar.highlight_map.lock() =
1383 HighlightMap::new(highlights_query.capture_names(), theme);
1384 }
1385 }
1386 }
1387
1388 pub fn grammar(&self) -> Option<&Arc<Grammar>> {
1389 self.grammar.as_ref()
1390 }
1391
1392 pub fn default_scope(self: &Arc<Self>) -> LanguageScope {
1393 LanguageScope {
1394 language: self.clone(),
1395 override_id: None,
1396 }
1397 }
1398
1399 pub fn lsp_id(&self) -> String {
1400 self.config.name.lsp_id()
1401 }
1402
1403 pub fn prettier_parser_name(&self) -> Option<&str> {
1404 self.config.prettier_parser_name.as_deref()
1405 }
1406}
1407
1408impl LanguageScope {
1409 pub fn path_suffixes(&self) -> &[String] {
1410 &self.language.path_suffixes()
1411 }
1412
1413 pub fn language_name(&self) -> LanguageName {
1414 self.language.config.name.clone()
1415 }
1416
1417 pub fn collapsed_placeholder(&self) -> &str {
1418 self.language.config.collapsed_placeholder.as_ref()
1419 }
1420
1421 /// Returns line prefix that is inserted in e.g. line continuations or
1422 /// in `toggle comments` action.
1423 pub fn line_comment_prefixes(&self) -> &[Arc<str>] {
1424 Override::as_option(
1425 self.config_override().map(|o| &o.line_comments),
1426 Some(&self.language.config.line_comments),
1427 )
1428 .map_or(&[] as &[_], |e| e.as_slice())
1429 }
1430
1431 pub fn block_comment_delimiters(&self) -> Option<(&Arc<str>, &Arc<str>)> {
1432 Override::as_option(
1433 self.config_override().map(|o| &o.block_comment),
1434 self.language.config.block_comment.as_ref(),
1435 )
1436 .map(|e| (&e.0, &e.1))
1437 }
1438
1439 /// Returns a list of language-specific word characters.
1440 ///
1441 /// By default, Zed treats alphanumeric characters (and '_') as word characters for
1442 /// the purpose of actions like 'move to next word end` or whole-word search.
1443 /// It additionally accounts for language's additional word characters.
1444 pub fn word_characters(&self) -> Option<&HashSet<char>> {
1445 Override::as_option(
1446 self.config_override().map(|o| &o.word_characters),
1447 Some(&self.language.config.word_characters),
1448 )
1449 }
1450
1451 /// Returns a list of bracket pairs for a given language with an additional
1452 /// piece of information about whether the particular bracket pair is currently active for a given language.
1453 pub fn brackets(&self) -> impl Iterator<Item = (&BracketPair, bool)> {
1454 let mut disabled_ids = self
1455 .config_override()
1456 .map_or(&[] as _, |o| o.disabled_bracket_ixs.as_slice());
1457 self.language
1458 .config
1459 .brackets
1460 .pairs
1461 .iter()
1462 .enumerate()
1463 .map(move |(ix, bracket)| {
1464 let mut is_enabled = true;
1465 if let Some(next_disabled_ix) = disabled_ids.first() {
1466 if ix == *next_disabled_ix as usize {
1467 disabled_ids = &disabled_ids[1..];
1468 is_enabled = false;
1469 }
1470 }
1471 (bracket, is_enabled)
1472 })
1473 }
1474
1475 pub fn should_autoclose_before(&self, c: char) -> bool {
1476 c.is_whitespace() || self.language.config.autoclose_before.contains(c)
1477 }
1478
1479 pub fn language_allowed(&self, name: &LanguageServerName) -> bool {
1480 let config = &self.language.config;
1481 let opt_in_servers = &config.scope_opt_in_language_servers;
1482 if opt_in_servers.iter().any(|o| *o == *name.0) {
1483 if let Some(over) = self.config_override() {
1484 over.opt_into_language_servers.iter().any(|o| *o == *name.0)
1485 } else {
1486 false
1487 }
1488 } else {
1489 true
1490 }
1491 }
1492
1493 pub fn override_name(&self) -> Option<&str> {
1494 let id = self.override_id?;
1495 let grammar = self.language.grammar.as_ref()?;
1496 let override_config = grammar.override_config.as_ref()?;
1497 override_config.values.get(&id).map(|e| e.0.as_str())
1498 }
1499
1500 fn config_override(&self) -> Option<&LanguageConfigOverride> {
1501 let id = self.override_id?;
1502 let grammar = self.language.grammar.as_ref()?;
1503 let override_config = grammar.override_config.as_ref()?;
1504 override_config.values.get(&id).map(|e| &e.1)
1505 }
1506}
1507
1508impl Hash for Language {
1509 fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
1510 self.id.hash(state)
1511 }
1512}
1513
1514impl PartialEq for Language {
1515 fn eq(&self, other: &Self) -> bool {
1516 self.id.eq(&other.id)
1517 }
1518}
1519
1520impl Eq for Language {}
1521
1522impl Debug for Language {
1523 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
1524 f.debug_struct("Language")
1525 .field("name", &self.config.name)
1526 .finish()
1527 }
1528}
1529
1530impl Grammar {
1531 pub fn id(&self) -> GrammarId {
1532 self.id
1533 }
1534
1535 fn parse_text(&self, text: &Rope, old_tree: Option<Tree>) -> Tree {
1536 with_parser(|parser| {
1537 parser
1538 .set_language(&self.ts_language)
1539 .expect("incompatible grammar");
1540 let mut chunks = text.chunks_in_range(0..text.len());
1541 parser
1542 .parse_with(
1543 &mut move |offset, _| {
1544 chunks.seek(offset);
1545 chunks.next().unwrap_or("").as_bytes()
1546 },
1547 old_tree.as_ref(),
1548 )
1549 .unwrap()
1550 })
1551 }
1552
1553 pub fn highlight_map(&self) -> HighlightMap {
1554 self.highlight_map.lock().clone()
1555 }
1556
1557 pub fn highlight_id_for_name(&self, name: &str) -> Option<HighlightId> {
1558 let capture_id = self
1559 .highlights_query
1560 .as_ref()?
1561 .capture_index_for_name(name)?;
1562 Some(self.highlight_map.lock().get(capture_id))
1563 }
1564}
1565
1566impl CodeLabel {
1567 pub fn plain(text: String, filter_text: Option<&str>) -> Self {
1568 let mut result = Self {
1569 runs: Vec::new(),
1570 filter_range: 0..text.len(),
1571 text,
1572 };
1573 if let Some(filter_text) = filter_text {
1574 if let Some(ix) = result.text.find(filter_text) {
1575 result.filter_range = ix..ix + filter_text.len();
1576 }
1577 }
1578 result
1579 }
1580
1581 pub fn push_str(&mut self, text: &str, highlight: Option<HighlightId>) {
1582 let start_ix = self.text.len();
1583 self.text.push_str(text);
1584 let end_ix = self.text.len();
1585 if let Some(highlight) = highlight {
1586 self.runs.push((start_ix..end_ix, highlight));
1587 }
1588 }
1589
1590 pub fn text(&self) -> &str {
1591 self.text.as_str()
1592 }
1593}
1594
1595impl From<String> for CodeLabel {
1596 fn from(value: String) -> Self {
1597 Self::plain(value, None)
1598 }
1599}
1600
1601impl From<&str> for CodeLabel {
1602 fn from(value: &str) -> Self {
1603 Self::plain(value.to_string(), None)
1604 }
1605}
1606
1607impl Ord for LanguageMatcher {
1608 fn cmp(&self, other: &Self) -> std::cmp::Ordering {
1609 self.path_suffixes.cmp(&other.path_suffixes).then_with(|| {
1610 self.first_line_pattern
1611 .as_ref()
1612 .map(Regex::as_str)
1613 .cmp(&other.first_line_pattern.as_ref().map(Regex::as_str))
1614 })
1615 }
1616}
1617
1618impl PartialOrd for LanguageMatcher {
1619 fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
1620 Some(self.cmp(other))
1621 }
1622}
1623
1624impl Eq for LanguageMatcher {}
1625
1626impl PartialEq for LanguageMatcher {
1627 fn eq(&self, other: &Self) -> bool {
1628 self.path_suffixes == other.path_suffixes
1629 && self.first_line_pattern.as_ref().map(Regex::as_str)
1630 == other.first_line_pattern.as_ref().map(Regex::as_str)
1631 }
1632}
1633
1634#[cfg(any(test, feature = "test-support"))]
1635impl Default for FakeLspAdapter {
1636 fn default() -> Self {
1637 Self {
1638 name: "the-fake-language-server",
1639 capabilities: lsp::LanguageServer::full_capabilities(),
1640 initializer: None,
1641 disk_based_diagnostics_progress_token: None,
1642 initialization_options: None,
1643 disk_based_diagnostics_sources: Vec::new(),
1644 prettier_plugins: Vec::new(),
1645 language_server_binary: LanguageServerBinary {
1646 path: "/the/fake/lsp/path".into(),
1647 arguments: vec![],
1648 env: Default::default(),
1649 },
1650 }
1651 }
1652}
1653
1654#[cfg(any(test, feature = "test-support"))]
1655#[async_trait(?Send)]
1656impl LspAdapter for FakeLspAdapter {
1657 fn name(&self) -> LanguageServerName {
1658 LanguageServerName(self.name.into())
1659 }
1660
1661 async fn check_if_user_installed(
1662 &self,
1663 _: &dyn LspAdapterDelegate,
1664 _: &AsyncAppContext,
1665 ) -> Option<LanguageServerBinary> {
1666 Some(self.language_server_binary.clone())
1667 }
1668
1669 fn get_language_server_command<'a>(
1670 self: Arc<Self>,
1671 _: Option<Arc<Path>>,
1672 _: Arc<dyn LspAdapterDelegate>,
1673 _: futures::lock::MutexGuard<'a, Option<LanguageServerBinary>>,
1674 _: &'a mut AsyncAppContext,
1675 ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
1676 async move { Ok(self.language_server_binary.clone()) }.boxed_local()
1677 }
1678
1679 async fn fetch_latest_server_version(
1680 &self,
1681 _: &dyn LspAdapterDelegate,
1682 ) -> Result<Box<dyn 'static + Send + Any>> {
1683 unreachable!();
1684 }
1685
1686 async fn fetch_server_binary(
1687 &self,
1688 _: Box<dyn 'static + Send + Any>,
1689 _: PathBuf,
1690 _: &dyn LspAdapterDelegate,
1691 ) -> Result<LanguageServerBinary> {
1692 unreachable!();
1693 }
1694
1695 async fn cached_server_binary(
1696 &self,
1697 _: PathBuf,
1698 _: &dyn LspAdapterDelegate,
1699 ) -> Option<LanguageServerBinary> {
1700 unreachable!();
1701 }
1702
1703 async fn installation_test_binary(&self, _: PathBuf) -> Option<LanguageServerBinary> {
1704 unreachable!();
1705 }
1706
1707 fn process_diagnostics(&self, _: &mut lsp::PublishDiagnosticsParams) {}
1708
1709 fn disk_based_diagnostic_sources(&self) -> Vec<String> {
1710 self.disk_based_diagnostics_sources.clone()
1711 }
1712
1713 fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
1714 self.disk_based_diagnostics_progress_token.clone()
1715 }
1716
1717 async fn initialization_options(
1718 self: Arc<Self>,
1719 _: &Arc<dyn LspAdapterDelegate>,
1720 ) -> Result<Option<Value>> {
1721 Ok(self.initialization_options.clone())
1722 }
1723}
1724
1725fn get_capture_indices(query: &Query, captures: &mut [(&str, &mut Option<u32>)]) {
1726 for (ix, name) in query.capture_names().iter().enumerate() {
1727 for (capture_name, index) in captures.iter_mut() {
1728 if capture_name == name {
1729 **index = Some(ix as u32);
1730 break;
1731 }
1732 }
1733 }
1734}
1735
1736pub fn point_to_lsp(point: PointUtf16) -> lsp::Position {
1737 lsp::Position::new(point.row, point.column)
1738}
1739
1740pub fn point_from_lsp(point: lsp::Position) -> Unclipped<PointUtf16> {
1741 Unclipped(PointUtf16::new(point.line, point.character))
1742}
1743
1744pub fn range_to_lsp(range: Range<PointUtf16>) -> lsp::Range {
1745 lsp::Range {
1746 start: point_to_lsp(range.start),
1747 end: point_to_lsp(range.end),
1748 }
1749}
1750
1751pub fn range_from_lsp(range: lsp::Range) -> Range<Unclipped<PointUtf16>> {
1752 let mut start = point_from_lsp(range.start);
1753 let mut end = point_from_lsp(range.end);
1754 if start > end {
1755 mem::swap(&mut start, &mut end);
1756 }
1757 start..end
1758}
1759
1760#[cfg(test)]
1761mod tests {
1762 use super::*;
1763 use gpui::TestAppContext;
1764
1765 #[gpui::test(iterations = 10)]
1766 async fn test_language_loading(cx: &mut TestAppContext) {
1767 let languages = LanguageRegistry::test(cx.executor());
1768 let languages = Arc::new(languages);
1769 languages.register_native_grammars([
1770 ("json", tree_sitter_json::LANGUAGE),
1771 ("rust", tree_sitter_rust::LANGUAGE),
1772 ]);
1773 languages.register_test_language(LanguageConfig {
1774 name: "JSON".into(),
1775 grammar: Some("json".into()),
1776 matcher: LanguageMatcher {
1777 path_suffixes: vec!["json".into()],
1778 ..Default::default()
1779 },
1780 ..Default::default()
1781 });
1782 languages.register_test_language(LanguageConfig {
1783 name: "Rust".into(),
1784 grammar: Some("rust".into()),
1785 matcher: LanguageMatcher {
1786 path_suffixes: vec!["rs".into()],
1787 ..Default::default()
1788 },
1789 ..Default::default()
1790 });
1791 assert_eq!(
1792 languages.language_names(),
1793 &[
1794 "JSON".to_string(),
1795 "Plain Text".to_string(),
1796 "Rust".to_string(),
1797 ]
1798 );
1799
1800 let rust1 = languages.language_for_name("Rust");
1801 let rust2 = languages.language_for_name("Rust");
1802
1803 // Ensure language is still listed even if it's being loaded.
1804 assert_eq!(
1805 languages.language_names(),
1806 &[
1807 "JSON".to_string(),
1808 "Plain Text".to_string(),
1809 "Rust".to_string(),
1810 ]
1811 );
1812
1813 let (rust1, rust2) = futures::join!(rust1, rust2);
1814 assert!(Arc::ptr_eq(&rust1.unwrap(), &rust2.unwrap()));
1815
1816 // Ensure language is still listed even after loading it.
1817 assert_eq!(
1818 languages.language_names(),
1819 &[
1820 "JSON".to_string(),
1821 "Plain Text".to_string(),
1822 "Rust".to_string(),
1823 ]
1824 );
1825
1826 // Loading an unknown language returns an error.
1827 assert!(languages.language_for_name("Unknown").await.is_err());
1828 }
1829}